• Mississippi River;
  • flood hydrology;
  • hydrologic measurements


This study tests the hypothesis that historical float-based discharge measurements on the Mississippi River systematically over-stated actual flood flows by 10% to > 30% relative to measurements using current meters. This assertion has been repeated over the past 25 years and recently has been used to adjust historical discharges used for flood-frequency analysis. This study tests the hypothesis above using 2150 historical discharge measurements digitized from the three principal gauging stations on the Middle Mississippi River (MMR): data that include 626 float-based discharges and 1516 meter-based discharges, including 122 paired measurements. Multiple comparative tests show that the hypothesis above cannot be supported; if anything, the float-based measurements slightly underestimate flows (not over-estimate) over a broad range of discharges up to large floods. In response to the purported data bias above (‘changing history’; Dieckmann RJ, Dyhouse GR. 1998. Changing history at St. Louis—adjusting historic flows for frequency analysis. First Federal Inter-Agency Hydrologic Modeling Conference, April 20–22, 1998. Las Vegas, NV; 4·31–4·36), historical flood discharges on the MMR have been modified, most by 10–20% and several by > 30%. These altered discharges are now being promulgated, in particular, through the Upper Mississippi River System Flow Frequency Study (UMRSFFS). New flow frequencies, flood profiles, and new flood maps from the UMRSFFS may significantly underestimate the actual flood hazard on the MMR if the original hydrologic data have been erroneously altered on the basis of an assumption of data bias. Copyright © 2010 John Wiley & Sons, Ltd.